The PHRASE COMBINER

The PHRASE COMBINER attempts to simplify the job of the final domain pattern recognizer by combining phrases from the initial parse into larger phrases whenever this is feasible. This combination takes place in a hierarchy of stages, so that various combination operations can be prioritized. For example, the attachment of certain prepositional phrases is performed before conjunction combination, so conjunction can apply to noun groups with prepositional phrases attached.

Each level of the phrase combination phase has two subphases: a defeat subphase, and a pattern matching subphase. If a pattern in the defeat subphase matches the input, then that string is prevented from matching any pattern in the matching subphase. For example, in general, ``for'' and ``of'' prepositions attach almost always to their closest noun group. These attachments are routinely made except in two cases: (1) there is a verb that explicitly subcategorizes for a ``for'' or ``of'' complement, or (2) the subject and object of the preposition form an important domain pattern that is recognized during the next phase (e.g. ``the production of golf clubs''). In these cases, defeat patterns are written to match the input and prevent the pp attachment rule from operating.

The PHRASE COMBINER performs the following tasks:

• Adjacent location NGs are merged when the result of the merger is consistent with the information in the gazetteer (e.g. ``Palo Alto, California'').
• Noun groups with company words as heads are combined with appositives, genitives, and prepositions to provide further information about the entity (e.g. ``Foobarco, the California company,'' ``Japan's Kobe Steel,'' or ``Aerospatiale of France.'') If any of the company names in the input are only possible companies (like Foobarco), matching one of these patterns will cause the possible company to recognized as a company for the duration of the text.
• Appositives and prepositions that associate people with titles and companies are combined, and their semantics processed. (e.g. ``John Smith, president and CEO of Foobarco'') Any possible company that matches the pattern is promoted to an actual company.
• Conjunctions of company names are combined (e.g. IBM, General Motors, and Foobarco) Again, possible companies can be promoted if they match the pattern.
• Certain patterns that can reliably be used to promote possible companies to actual companies are recognized, even though they don't directly contribute any information to template slots (e.g. ``the board of directors of Foobarco'')
• Conjoined verb groups are recognized, as well as certain phrases that can be treated by subsequent analysis as complex verb groups (e.g. ``manufacture and market'', ``planning to set up'', ``announced a plan to form'').
• Finally, ``of'' and ``for'' prepositions are attached to their adjacent noun groups, and conjoined noun groups are combined, unless defeated by defeat patterns.

As an example of the operation of the PHRASE COMBINER, consider the system's processing of the second sentence of the walkthrough text:

CN: "JOINT-VENTURE BRIDGESTONE " (0,5) Head: BRIDGESTONE
NG: "SPORTS " (5,6) Head: SPORTS
LOC: "TAIWAN " (6,7) Head: TAIWAN
ACTIVE/PASSIVE: "CAPITALIZED " (9,10) Head: CAPITALIZED
PREP: "AT " (10,11) Head: AT
NG: "20 MILLION NEW TAIWAN DOLLARS " (11,16) Head: DOLLARS
ACTIVE: "WILL START " (17,19) Head: START
NG: "PRODUCTION " (19,20) Head: PRODUCTION
PREP: "IN " (20,21) Head: IN
NG: "JANUARY 1990 " (21,23) Head: -DATE-
PREP: "WITH " (23,24) Head: WITH
NG: "PRODUCTION " (24,25) Head: PRODUCTION
PREP: "OF " (25,26) Head: OF
NG: "20000 IRON AND METAL WOOD CLUBS " (26,32) Head: CLUBS
NG: "MONTH " (32,34) Head: MONTH

The PHRASE COMBINER combined two noun groups in this sentence. The combination producing ``20000 IRON AND METAL WOOD CLUBS'' was correct. Unfortunately, because of the problem cited above with the word ``SPORTS,'' the system did not correctly recognize the joint venture company name, and the combiner formed the appositive ``JOINT VENTURE BRIDGESTONE'' and assigned BRIDGESTONE the role as the joint venture company. Of course, BRIDGESTONE was already identified as one of the parent entities, so this was the source of the mistake that led to discarding the entire analysis of this text. Also, because the previous sentence said that the joint venture was ``in Taiwan'' we identified Taiwan as the location of Bridgestone.